the income inequality of france in historical...

European Review of Economic History, 4, 59-83. Printed in the United Kingdom © 2000 Cambridge University Press

The income inequality of France inhistorical perspectiveCHRISTIAN MORRISSONf AND WAYNE SNYDER^f Universite de Paris I, 36 chemin Desvallieres, 92410 Ville d'Avray, France.% 5, rue des Chaudronniers, 13100 Aix-en-Provence, France.

France presents an unusual case because, unlike several other Europeancountries, there are no estimates of the income distribution for theeighteenth and nineteenth centuries. This is a serious deficiency becauseit limits the ability to understand how an important dimension of thesocio-economic fabric changed during the years preceding and coincidingwith the beginning of France's industrial development. In this article weprovide estimates that use tax data and the social tables of the eighteenthand nineteenth centuries. While this data does not provide a basis for aperfectly accurate assessment of the income distribution, it does permitan evaluation of the general magnitude of inequality and how it varied inthat period. The results suggest that inequality during the eighteenthcentury was large but decreased during the revolutionary period(1790-1815). Afterwards, and in accordance with Kuznets' hypothesis,when industrialisation began about 1830, inequality increased untilsometime in the 1860s when it began its slow decline towards greaterequality that characterises the twentieth century.

1. Introduction

Shortly after the 1949 creation of the OECD (OEEC at the time), membercountries agreed to prepare national accounts on a standardised basis.Today a continuous and reliable series of national accounts is available formost member countries. Several countries undertook to prepare accounts oftheir income distributions as well. For more than a decade the World Bank(see World Bank 1998) has collected and published income distributions foras many countries worldwide as possible. In France, INSEE has beenresponsible for both sets of accounts and as a result consistent and reliabledata have been published since the 1950s.1 However, there are no reliableestimates of how income was distributed before World War II, althoughToutain (1987) used production data to estimate 10-year GDP averages forthe period 1789 to 1982. The absence of information about the income

1 Income distribution data are published at irregular intervals in the INSEE series: 'Lesrevenus flscaux des menages'.

6o European Review of Economic History

distribution in the eighteenth and nineteenth centuries presents a seriousimpediment to understanding how the socioeconomic fabric of Franceevolved during those times that preceded, and then coincided with, thebeginnings of industrial development. This article presents our effort toreduce the deficiency of information about the income distribution ofFrance during the eighteenth and nineteenth centuries.

Data about individual or household incomes during the eighteenth andnineteenth centuries are a rare commodity in most countries, and in Francetheir absence is particularly notable. In making estimates of the income dis-tribution we have relied on a variety of proxy sources, none of which havethe same consistency or reliability as those that are available for many coun-tries in the second half of the twentieth century. Where possible we haveprovided sensitivity tests to indicate how our estimates would change if theunderlying assumptions about the data were different. Given the limitedextent and uncertain quality of the sources and in light of the researchundertaken elsewhere, our efforts to establish certain orders of magnitudefor the income distribution might be viewed as overly audacious. Our hopeis that this initial probing will induce others to seek new sources capable ofincreasing our knowledge of the income distribution in France during theeighteenth and nineteenth centuries.

2. The search for income proxies

There are no completely reliable data about income in France until afterWorld War II. The income tax came into effect during World War I, but thenumber of persons subject to the tax was so restricted that it provides onlylimited knowledge about the distribution of income. Toutain's GDP esti-mates for the nineteenth century provide little information that is directlyrelevant to determining how income might have been distributed, althoughthe information can be used to corroborate income estimates based on otherinformation. In the absence of direct measures of income, we have usedtaxes that imperfectly reflect the incomes of those assessed as well as thesocial tables of contemporaries who had little recourse to the sources ofnational statistics as they are known today. These proxy sources areundoubtedly imperfect indicators of income, but together they provide aconsistency check and a means for evaluating how sensitive the data are tochanges in the estimates. The case of France may be qualitatively differentfrom some other European countries, but the task of finding proxy data thataccurately reflect the income distribution has often confronted economichistorians.

Soltow (1987) used a federal housing survey to estimate the distributionof income for the United States in 1798 because there was a general beliefat the time that house values were proportional to the ability to pay. Evenso, the data present a problem because the number of families living in each

The income inequality of France in historical perspective 61

dwelling is unknown. Soltow (1989) was able to make an estimate of theincome distribution for the inhabitants of Amsterdam as early as 1585.However, this and his later estimates for Amsterdam are based on taxes thatare simply presumed to be based on general earning-power. A further limi-tation is that his estimates are based on tax rolls that excluded 50 per centor more of the population, mainly those in the lower occupational cat-egories. Dumke's (1991) comprehensive survey of the multiple efforts todetermine the historical evolution of income inequality in Germany (andpreviously Prussia) illustrate the need to rely on proxy data for which thequality is uncertain and variable. Among those who have focused on the his-torical evolution of inequality in Britain are Lindert (1980, 1986, 1994,1999) and Williamson (1979, 1985). Like others they have sometimes usedtaxes as a proxy for income, but they have also relied on social tables for theanalysis of income (Lindert and Williamson 1983). Relying on proxymeasures of income is always open to questioning, and Feinstein (1988) hasasserted that Williamson's (1985) conclusions about the evolution ofinequality in the nineteenth century rests on assumptions about the under-lying data that are questionable and in some instances incorrect.

3. The eighteenth century

In the absence of any direct measure of income in eighteenth centuryFrance, we present estimates of the income distribution based on three sep-arate methods: (1) inferences from impositions of the capitation tax, (2)analysis of the socio-professional structure, and (3) an eighteenth centuryeffort to quantify the economic structure of France.

3.1. The capitation tax

The capitation tax, introduced by Louis XIV in 1695 to help finance hiswars, was the first tax intended to be universal. Except for 1699 and 1700,it was collected every year up to the Revolution. The capitation was notFrance's first direct tax. Another direct tax, the taille, had been introducedin 1449 and remained a major source of royal income until the Revolution.But it has major deficiencies as a proxy for income. In some regions of thecountry the taille was based on an estimate of a person's income {la taille per-sonnelle), but elsewhere only land was taxed (la taille reelle). Furthermore,most nobles were exempt as were certain professions and even entire towns.Without touching the privilege of those exempt, after 1733 a schedule ofrates, the taille tarifee3 applicable for all types of income was established, butit was effective in only limited areas because it depended on the completionof a national cadastre that was beyond the capacity of the administration(Touzery 1994, pp. 73-86; Bonney 1999). During the eighteenth centuryOther direct taxes were introduced (dixieme, cinquieme and vingtieme), but

62 European Review of Economic History

exemptions were so numerous as to render none of them suitable as a proxyfor income (Bonney 1995, pp. 472-88).

The capitation is no panacea for eliminating the deficiencies of the otherdirect taxes as a proxy for income, but it provides the most reliable meansof estimating how income was distributed in the eighteenth century. Thetax, as applied, was only roughly related to a person's income and some per-sons were exempt. In the first year the population was divided into 22 socialclasses; within each class persons were assessed identically.2 The onlyintended exemptions were people declared by their parish priest to be toopoor to pay. Almost immediately, however, the church negotiated an annualpayment, and some nobles and a few others were exempt because of theroyal offices they held. The king's administrators soon found it easier toassign each community a lump sum (impot de repartition) rather than dependon whatever the community might collect based on the original tax sched-ule (impot de quotite). At the same time and at their own initiative, com-munities began to divide their assigned tax by allocating shares to heads ofhouseholds according to their income rather than blindly applying identicaltaxes to persons in the same profession but who might have widely diver-gent incomes. The average level of assessment varies by region and also asbetween the larger cities and the more numerous villages. The records showthat nobles and other privileged persons were more successful than othersin having their assessments reduced, but taxpayers were generally able tocompare their assessments to those of others and when they felt theirburden was unfair, they would cry for justice (Kwass 1994, p. 147). Theprolific expert on Old Regime taxes and finances, Marion, concluded that'the capitation was a genuine tax on gross income as estimated by the admin-istration' (Marion 1910, p. 24). The tax rolls themselves, however, containfew examples of income to compare with the tax assessments. In the few taxrolls that have both, the range of tax rate percentages is narrow and in gen-eral suggest that nearly flat rates were used. Thus, the capitation appears tohave evolved towards a flat rate income tax but whose universality hadmajor shortcomings.

At the very bottom of the social hierarchy it is uncertain how many peoplewere exempt because they had no income at all or their income was so smallthey were excused from the tax. Many communities listed their poor in thecapitation rolls, but it is impossible to ascertain whether or not the rollswere comprehensive. During the Revolutionary period the administrationestablished a Comite de Mendicite to estimate the extent of poverty inFrance. Their report concluded that as much as 10 per cent of the popu-lation required outside assistance to maintain life and limb (Hufton 1974).In our survey of the capitation rolls, we found 4.3 per cent who were

2 Two of the most useful analyses of the capitation tax as it was originally conceived areprovided by Bluche and Solnon (1983) and Guery (1986).


assessed nothing because they were either too poor or invalid. Many of thesewould have been single persons, but undoubtedly some had wives and chil-dren. In addition, vagabonds and beggars were usually omitted from thecapitation rolls. Together with the other poor and invalid, the percentage ofpersons at the bottom of the income distribution was probably close to the10 per cent reported by the Revolutionary committee.

Further up the income scale, throughout the eighteenth century 75 percent of the population of France worked in the agricultural sector. Theimportance of the own-consumption of food was undoubtedly greatest forthe lower level workers, but there is no way to ascertain from the taxassessments whether or not this was considered when the taxes wereassigned. Based on the capitation statistics, about 5 per cent of the popu-lation in eighteenth century France were engaged as household servants,most of whom were housed and fed at their place of work. The scale of tax-ation for servants was similar throughout France and remained more or lessconstant throughout the eighteenth century. The tax rates for the higherranked servants were two or three times the rate for a simple domestique andwomen were normally assessed less than males. By themselves the taxassessments are an inaccurate indicator of their income because they dis-regard the income in-kind that servants received in the form of lodging andmeals.

For nobles who were exempt from the capitation, we were usually able tolearn the amount that would have been assessed because the assessment waslisted and crossed out by the word 'exempt.' But there are cases where noassessment was indicated, and these contribute to underestimating thedegree of inequality in the capitation rolls. When persons had multipleincome sources, the capitation law provided that they should be taxed onlyfor their largest source of income. As many nobles and bourgeois had severalsources of income, based on land and stipends associated with official func-tions, the capitation underestimates their total income. Unfortunately thecapitation rolls provide no information about the nature or extent of anyincome that escaped taxation. The capitation provided that if individualfamily members had separate sources of income, each would be taxedaccordingly. This provision affected principally the families of noblesbecause they were the ones where husbands, wives, and often children aswell, had separate and taxable income.

Given these caveats about using the capitation to approximate incomeduring the eighteenth century, let us look at how the capitation assessmentswere distributed. We looked for capitation rolls in the archives of everydepartment in France. Over a period of three years we collected randomsamples from 189 rolls that we found in the archives of 56 departments.

It seems reasonable to presume that the many rolls that have disappearedwere lost due to fire, wars, vandalism and simple negligence, leaving theremaining rolls a random sample of the original population. The number of


Table I. Unadjusted distribution of capitation assessments (percentageof assessments by group of taxpayers).

Group

Top decile9th decile4th quintile3rd quintile1st and 2nd quintile

1695-1704

661311

64

1705-47

61

1413

66

1748-59

49181610

7

1760-90

6016

1365

Note: Category: Taxpayers.Source: Capitation rolls from 56 departements (304,167 taxpayers).

persons assessed in these rolls amounts to 304,167. Since only heads-of-households were assessed, the total number of persons represented by therolls is 1,300,000. Table 1 shows the assessments by taxpayer group. Withthe exception of the period 1741-73, the distribution does not appear tohave changed significantly during the eighteenth century. The top decile ofthe population were taxed about 60 to 65 per cent of all the assessmentswhile the bottom 40 per cent were taxed about four to five per cent of totalassessments. The period 1741-73 was influenced by the introduction ofanother tax, the vingtieme. It may be that the reduced capitation assess-ments of this period were the result of reductions accorded to people withbusiness income since they were the ones who were most affected by thenew tax.

The next task is to adjust the data of Table 1 to account for the own-con-sumption of persons in the agricultural sector, the income in-kind receivedby servants, and the fact that some noble and bourgeois families hadadditional sources of income that were untaxed.3 We have presumed thatservants and the lower ranks of those in agriculture received about half theirincome as income in kind. As for the untaxed income of nobles and bour-geois^, we assumed this was no more than five per cent of their total income.These adjustments are designed to modify the distribution of assessmentsso they more accurately reflect the distribution of income. The result ofthese adjustments are given in Table 2 and can be compared to the originalcapitation data of Table 1.

Two observations may be made about the distribution of income shownin Table 2. First, the net effect of the adjustments reduces the share ofincome attributed to the top two quintiles while increasing the income forthe bottom three quintiles, due principally to the importance of income inkind received by persons in the latter groups. The adjustments undoubtedly

3 The fact that some nobles and bourgeois managed to have their taxes reduced in greaterproportion than others need not concern us here as it is the assessments themselves thatare the better estimate of income.


Table 2. Distribution of revenue based on capitation assessments:adjusted for income-in-kind and own-consumption (percentage ofassessments by taxpayer groups).

Group

Top decile9th decile4th quintile3rd quintile1st and 2nd quintile

1695-1704

621210

88

1705-47

571312

810

1748-59

4617151111

1760-90

561512

89

Note: Category: Taxpayers. Data adjustments: corrected for income-in-kind and auto-consumption based on the assumption that 50 per cent of the income of quintiles 1 and 2and 33 per cent of the income of quintile 3 are received as income-in-kind and auto-consumption that is excluded from the data of Table 1.Source: Capitation roles from 56 departements (304,167 taxpayers).

contain immeasurable errors. If the error was as large as ±10 per cent, thenthe range for the top decile during the period 1760-90 would be 51 per cent<56 per cent <6i per cent. Second, there may have been a modest decreasein inequality during the eighteenth century. We will return to this issue laterafter looking at other evidence about the income distribution at the end ofthe eighteenth century.

3.2. The socio-professional structure

Estimates of the socio-professional population exist for the years 1765 and1788. However, the revenue data for the various categories are more reliablefor 1788 than for 1765. Expilly's (1780) estimate of the socio-professionalstructure appears to be the best available for France in the late eighteenthcentury. We have assumed that most heads-of-household were in four-person families, except for non-agricultural workers whose families hadthree persons, and we have assumed that most servants were single.4 Thenumber of nobles and clergy is open to debate and the same can be said forthe number of bourgeois. The information about persons in these groups isnot sufficiently reliable to estimate the number of active persons in eachgroup or their incomes. We provide only estimates of the total number ofnobles, clergy and bourgeois, and the combined income for these threegroups. According to Toutain (1987, p. 56) the GDP for France during thedecade 1781-1790 was 5,941m francs expressed in current prices. We have

4 In truth, there is no precise figure for household size in the eighteenth century. Becauseof the disturbances and the great famine of 1709-10, the size was different at differenttimes. Dupaquier (1991, p. 55) notes that while Des Cilleuls (1885) used 4.5 persons perfeux in estimating the eighteenth century population of France, he believed it is moreprudent to use a coefficient of 4 due to the various crises.


Table 3. Income distribution by social group in 1788.

Group

(1) Nobles and clergy(2) Bourgeois(3) Shopkeepers and

artisans(4) Workers (non-agriculture)(5) Servants (non-agriculture)(6a) Small scale farmers(6b) Large scale farmers(7) Agriculture: day labourers

and servants(8) Mixed workers

(agriculture and industry)8

Total

Population(thousands)

5402,160

3,2401,5001,0805,2502,250

10,150

1,80028,000

Income perhousehold(livres)

6002 0 0

100

250880

160

300

Total income(millions of livres)'high'

}i>955

486100

100

330494

400

1354,000

'low'

>>| I J 7 I 5

486115

115390584

460

1354,000

Notes: Category: Population. Agricultural workers who spent part-time as industrialworkers, many for cottage industry textiles, others as assistants to artisans. Dataadjustments: based on the assumption of 1 male worker for every 4 persons except forworkers in the non-agricultural sector (i for 3 persons) and town servants (1 for 1.08persons). The 'low' estimate is based on the assumption that the share of agriculturalincome going to groups 1 and 2 is only 25 per cent instead of 33 per cent. This results inan increase of the income share going to small and large farmers. The 'low' estimate alsoassumes that the income of agricultural day labourers, servants and non-agriculturalworkers is increased by 15 per cent which further reduces the income of groups 1 and 2.Sources: Population: Braudel and Labrousse (1970, pp. 493, 607, and 659) and Expilly(1780); Income: Braudel and Labrousse (1970), Peuchet (1805, p. 391), and Weir (1991).

used a lower figure of 4,000111 francs for the income for 1788 because weexcluded the income and production of the State, and because Toutain mayhave overvalued agricultural income. With these hypotheses we can obtain anapproximation of how income in 1788 may have been distributed among thevarious socio-professional classes; these estimates are provided in Table 3.

Because the income of nobles, clergy and bourgeois is less well known thanfor the other groups. Table 3 provides two estimates (high and low) to illus-trate how sensitive the income distribution is to alternative hypotheses. The'low' estimate incorporates the results of decreasing the share of agriculturalincome going to these groups from 33 per cent to 25 per cent whichincreases the incomes of the small and large scale farmers. The 'low' esti-mate also includes a small increase of the incomes of agricultural daylabourers and servants3 and non-agricultural workers, that further decreasesthe income attributed to nobles, clergy and bourgeois.

The estimates of Table 3 can be transformed into an income distribution,


Table 4. Household income distribution in 1788 (per cent).

Groups Population^ Totalincome^'high'

Adjusteddistributions

Population Income'High' 'Low'

I. Groups 1 and 2 9.6II. Groups 3 and 6b 19.6III. Groups 4J 5, 6a, 7 and 8 70.8

48.824.526.7

102070

532126

472330

Notes: Category: Population. Data adjustments: adjusted distributions are from Table 3.Sources: Population: Braudel and Labrousse (1970, pp. 493, 607, and 659) and Expilly(1780); Income: Braudel and Labrousse (1970), Peuchet (1805, p. 391), and Weir (1991).

but one that can provide only a rough approximation of what a moredetailed distribution might give. This is shown in Table 4.

The incomes taken from Table 3 and combined as groups I, II and III inTable 4 have some deficiencies because the three categories are not entirelyhomogeneous. There are large differences in the incomes of persons ingroups 1 and 2 of Table 3 (nobles, clergy and bourgeois). The wealthiest wereincredibly rich, even by today's standards. But within these groups therewere also nobles and clergy who were as poor as the proverbial church mice.Many of the clergy and some nobles belong in group II of Table 4. By thesame token, some of the merchants and large scale farmers included ingroup II undoubtedly had incomes that would place them in group I. Someof the persons in group II whose incomes were far below the average shouldreally be included in group III. We have adjusted the 'raw data' from Table3 to reflect the reallocation of persons among the three groups of Table 4.The difference between it and the data from Table 1 is not huge but becauseit is based on a more explicit analysis of the socio-professional structure, theresults presented in Table 4 may have greater accuracy in showing how theincome was distributed in 1788 than those of Table 2. The 'adjusted distri-bution' of Table 4 illustrates the sensitivity of the estimates to variants in theunderlying assumptions. The 'high' estimate attributes 53 per cent of theincome to the top 10 per cent of the population whereas the 'low' estimateattributes 45 per cent. We believe the lower figure does not represent areasonable estimate because the measurement errors most likely are biasedtoward decreasing the income attributed to the top 10 per cent.Consequently, it appears to us that the minimum percentage of incomereceived by the top 10 per cent was probably not less than 50 per cent.

3.3. An eighteenth century estimate

As early as 1694 Vauban, the renowned military architect, had attempted toelaborate the socio-professional structure of France in order to illustrate


( I )(2)

(3)(4)(5)(6)(7)(8)Total

Table 5. Isnard'sincome distribution: 1781.

Heads-of-households

(thousands)

3353 0 0

4 0 0

6 0 0

8 0 0

1,0001,2001,4006,035

Note: Category: Households.Source: Isnard (1781).

(per cent)

5-55.06.6

10.013-216.619.923.2

100.0

Income perhousehold

Total[ income

(livres) (millions of livres) (per cent)

3,670 1.1,8001,200

8 0 0

5 0 0

4 0 0

3 0 0

2 0 0

4:

,2305404804804 0 0

4 0 0

3602 8 0

,170

29.512.911.511.5

9.69.68.66.7

100.0

how a flat rate tax on income could solve Louis XTV's revenue problems.5

Nearly a century later, Thorillon (1787) published another illustration of therevenue that a flat rate tax on income might be capable of producing. Aboutthe same time a third estimate of the income of France was prepared byRenard (1790) for the same purpose. Neither of the late eighteenth centuryestimates of income have any great credibility. We have not used any ofthese early attempts to estimate social tables for France because all havesignificant deficiencies.

Isnard (1781), an engineer cum economist, produced another more coher-ent estimate. Like the others, his purpose was to illustrate the revenue thatcould be obtained from a flat tax of 10 per cent on income. His estimatesabout the population and its income appear more realistic than those pro-duced by the other authors and closely approximate figures accepted todayfor the closing decades of the eighteenth century. The details of hisestimates are presented in Table 5.

It is remarkable that Isnard's estimates of the income structure, based onsuch a limited knowledge of France's eighteenth century economy corre-spond closely with the best informed estimates of that period that we havetoday. His estimate of the number of heads of households implies a popu-lation of about 24m, perhaps 4m less than is generally accepted to have beenFrance's population toward the end of the eighteenth century. It is also clearthat his underestimate of France's population mainly resulted from anunderestimate of both the number of persons in the lowest income categoryand the existence of a significant number of poor who had no income at alland depended on the charity of others. Given that Isnard miscalculated the

5 Vauban's proposal is reproduced in Boislisle (1874.3 pp. 561-74).

5th quintile4th quintile3rd quintile2nd quintile^1st quintile /

7 i12

8

9


Table 6. Income distribution: late eighteenth century.

Income group Adjusted Table 4 Isnard's Authors' proposed estimatescapitation groups unadjustedassessments estimates 'high' 'low'

66 6016 16

8 10

I 2

10th decile 56 53 43 52 478th and 9th decile 21 21 24 24 231st to 7th decile 23 26 33 24 30

Note: Category: Population.Source: Tables 2, 4 and 5.

number of persons in the lowest income groups, we have adjusted hisdata to correct for the underestimation. These estimates are summarisedin Table 6 along with our proposals which include a 'high' and a 'low'estimate that we believe incorporate the most accurate parts from the otherdistributions.

It seems likely that the capitation data tended to underestimate theincome of persons in the bottom 40 per cent whereas Isnard's omission ofa large number of poor overestimated the income of the bottom 40 percent. Adjusting for these differences, we arrive at 51-53 per cent for thetenth decile that is somewhat below what the capitation data suggest andsomewhat higher than we derived from the group data of Table 4, andconsiderably more than is contained in Isnard's unadjusted estimates. Theadjusted capitation data of Table 2 along with the other estimates of Table6 suggest that the level of inequality in the years leading up to theRevolution was probably little different from what it had been nearly acentury earlier.

Since all the estimates of Table 6 are subject to error, it is useful to exam-ine how sensitive they are to change. Rather than examining the sensitivityof each distribution separately, we limit our discussion to changes in onlyour proposed 'high' and 'low' estimates. The 'low' estimate reduces the topquintile from 66 per cent to 60 per cent and raises the income attributed tothe lower groups. For example, the two lowest quintiles would receive 14per cent as compared with the 10 per cent according to the 'high' estimate.The potential consequences of such changes for the Gini coefficient(measure of inequality) are discussed below.

Our proposed synthesis of the income distribution for France in the lateeighteenth century (Table 6) produces a Gini coefficient of 0.59. It is usefulto examine how this inequality compares with estimates made for other


countries during similar periods. Soltow (1989, p. 78) estimated a Gini of0.69 for Amsterdam in 1742. Clark (1951) estimated that the Gini forPrussia in 1852 was 0.34. More recently Dumke (1991, p. 129) has re-esti-mated it as 0.36 in 1850. Lindert (1999) estimated a Gini of 0.52 for Britainin 1759 and 0.59 for 1801. Lindert and Williamson (1983, p. 98-102) esti-mated that in the period 1801-3 the Gini coefficient was 0.58; they also citeearlier estimates that put the Gini in the range 0.51-0.58. All the estimates,except for Amsterdam, are somewhat - but not dramatically - lower thanour estimate for France. Lindert (1986, p. 1149) estimated that the share ofpre-tax income received by the top 10 per cent of households in Britainamounted to 49 per cent in 1803. This is marginally less than the 51-55 percent we propose for late eighteenth century France. The Gini coefficient forBritain is particularly interesting because its industrial revolution began twoor three decades before France, so the figure for 1759 is likely to be com-parable with France's development in 1788, and as such it suggests thatincome inequality in Britain may have been somewhat less than in France.The estimated Gini coefficient for France in the late eighteenth centurycorresponds fairly closely to the estimates of income inequality that existedduring the 1960s in such Third World countries as Brazil, Kenya, andMexico.

Our estimates appear to portray a level of inequality that was not greatlydifferent from elsewhere in Europe where comparable estimates exist.However, there is some uncertainty about this as both England and Prussiaappear to have had somewhat less inequality during roughly comparableperiods (Lindert 1999, Dumke 1991). Nevertheless it is useful to enquirehow sensitive these results are to changes in our estimates. We can investi-gate how changes in the income distribution would be reflected in changesin the measure of inequality, the Gini coefficient. Based on our 'proposeddistribution of income' we found a Gini coefficient of 0.59. When we recal-culate it based on the assumption that the income of the top quintile couldvary by ±10 per cent, the effect is to increase the Gini coefficient to 0.66 orto lower it to 0.55. The higher estimate would put inequality in France sub-stantially higher than elsewhere in Europe at about the same period. Thelower estimate would place France somewhat but not greatly below the esti-mates for Britain.

4. The nineteenth century

The French Revolution brought many changes to France that affected theincome distribution. Unfortunately there is a paucity of data that could beused as a proxy for income in order to evaluate the effects of these changes.In 1792 the capitation was abolished along with the entire tax system of theOld Regime. The new taxes that replaced the old ones were not sufficientlyuniversal to provide information about the entire socio-professional


structure. One third of the population was exempt from the widest basedtax, the impot immobilier, because they owned no property. There are noviable indicators that can be used to approximate how the distribution ofincome changed between 1790 and the 1830s.

While it is impossible to estimate the income distribution during thisforty year period, there are several changes that strongly suggest the direc-tion of change during this period. After the Revolution the lower classesbenefited from the abolishment of the dime, a tax that had affected themdisproportionately. The abolishment of the feudal rights of nobles alsoenhanced the income of the lower class of farmers who no longer were sub-ject to work corvees and other obligations to their seigneurs. Perhaps themost important aspect was the confiscation of church properties and thoseof many nobles. These were auctioned as biens nationaux. While the largestgroup of buyers were the larger farmers and bourgeois who already pos-sessed land and had the means to acquire more, there were many from thelower ranks of the agricultural hierarchy who were also able to acquiresome of the lands auctioned by the government. As a consequence, thestructure within the agricultural sector changed dramatically. In the newdepartment Nord, the share of land held by the clergy and nobles decreasedfrom 42 per cent in 1788 to 12 per cent in 1802, and the share held by thepaysans increased from 30 per cent to 42 per cent.6 Before the Revolutionfarmers in the villages of southeastern France who were deemed 'depen-dent' because they didn't control enough land and other resources to feedand house their families without seeking outside employment or whodepended in part on charity outnumbered by more than two to one thosewho were economically 'independent'. But after the Revolution the per-centage of 'dependents' decreased from 46 per cent to 38 per cent while thepercentage of 'independents' increased from 20 per cent to 32 per cent.7

Another related indication of the benefit that workers received relative tothe higher classes is that the salary of an urban worker increased by 62 percent from 1785-89 to 1797-1803 while the price of wheat (the best singleindicator of the cost of living) increased only 28 per cent during the sameperiod.

Postel-Vinay (1989) has shown how the inflation during theRevolutionary years benefited those agricultural classes (sharecroppers)who were able to use a depreciated currency to purchase fixed capital andanimals for production. Eventually the importance of in-kind sharecroppingdecreased and was replaced by monetary contracts that favoured the tenantfarmers who could pay their debts in a depreciated money. Anotherconsequence of the inflation was that persons in the higher income cat-egories who had paid dearly to obtain the privileged posts and their

6 Braudel and Labrousse (1976, p. 62).7 Snyder and Ostroot (forthcoming).


associated emoluments of the Old Regime found afterwards they had lostthe quasi-totality of their investment because they were reimbursed in adepreciated currency.

These changes clearly had the effect of increasing the share of incomegoing to the lower ranks while at the same time the share going to the highofficials of the church and many nobles decreased. Although the impact ofthese changes on income inequality cannot be quantified, the direction isclear. The degree of inequality that existed prior to the Revolution wasdecreased by changes that benefited the lower classes. It seems likely thatthis redressing of the income distribution continued from the last decade ofthe eighteenth century until about the third decade of the nineteenthcentury.

4.1. The evolution of inequality: 1831-66

The decline in inequality that began during the Revolutionary period mayhave lasted less than forty years. The turning point appears to haveoccurred about 1830, after which inequality continued to increase well intothe 1860s.

Beginning with the Napoleonic era, the administration took an increasinginterest in conducting censuses that attempted to describe the socioeco-nomic structure of France. Verley (1993) and Marchand and Thelot (1997)provide analyses of the censuses of 1831 and 1866. However, their presen-tations are not entirely suitable for the purpose of constructing income dis-tributions for these years. Their employment categories are tooheterogeneous to serve as income groups and the dispersion within some ofthe groups is large. For example, according to Chanut et al (1995) in1860-65, 25 Pe r c e n t of the workers in the industrial sector had less than 63per cent of the sector's average salary. The implication is that within theindustrial sector there was an elite of highly skilled workers whose salarieswere much higher than the large majority, perhaps equal to or larger thanthe income of small merchants. The same is true of other categories as well.We sub-divided several categories in order to arrive at more homogeneousgroups: farmers were divided between large and small according to the sizeof exploitation based on the analysis of Toutain (1992); the divisionbetween employers and the self-employed was based on the statistics of thepatent tax provided by Verley; and civil servants were separated accordingto their rank. These, together with information about their respectiveincomes, enable us to estimate both the number and income for the majoremployment categories that are given in Table 7.

Based on the data in Table 7, the increase in inequality that occurredbetween 1831 and 1866 can be evaluated in several ways. If we combine thethree highest income groups of Table 7, (1) large farmers, (2) employersand (3) high level civil servants, together these groups accounted for 10 per


Table 7. Employment

(1) Large farmersSmall farmersAgricultural workers andservants

(2) EmployersSelf-employedWhite collar employeesBlue collar employees

(3) High level civil servantsLow level civil servants

Total

and income: 1831

Employment(thousands)

1831

7744,753

4,3H510

2,0403 0 2

2,098

170170

15,131

1866

8 2 0

5,035

4,302

6532,614

8273,462

300300

18,313

and 1866.

Averageincome(francs)1831

1,725280

252

4,96o390513580

1,023347573

1866

3,301538

372

9,490582692800

1,387463

1,017

Total income(millionsof francs)1831

i,3351,335

1,0882,530

795154

1,218

17459

8,688

1866

2,7072,707

i,5996,2021,520

5722,771

416

13918,633

Notes: Category: Labour force. Data adjustments: the decomposition of the three groups ofrevenue uses the following information: (1) Based on Toutain (1992), we have supposedthat 50 per cent of agricultural income is received by 14 per cent of the farmers. (2) Basedon the patente statistics of employers and the self-employed provided by Verley (1993), 19per cent of taxpayers received 71 per cent of the profits; his study also provides estimates ofthe interest and property income received by this group.Sources: Verley (1993), Marchand and Thelot (1997), Zylberman (1969), and seediscussion in text.

cent of employment in both years and incorporate those persons whowould be included in the tenth decile of the income distribution. In 1831their combined income amounted to 46 per cent of total income whereasby 1866 it had increased to 50 per cent of total income, an indication of theincrease in inequality. If we look at the lowest income group that includesagricultural workers and both rural and urban servants, their numbersremained virtually the same in 1866 as they had been in 1831; the income ofthis group increased 46 per cent in nominal terms between 1831 and 1866whereas total income in France increased by 115 per cent, an indication ofthe deteriorating position of persons who would be in the first quintile of anincome distribution. Between 1831 and 1866 the average income of blue andwhite collar employees increased less than 40 per cent in nominal terms, butduring the same period the average income per person income increased by77 per cent. These persons and their income roughly occupy a position inthe income distribution that corresponds to the second quintile. The factthat the share of total income received by persons occupying the first andsecond quintiles declined while the share of income going to the top decileincreased is a clear indication of the increasing inequality that occurredbetween 1831 and 1866. One factor that slowed the rise of income for per-


Table 8. Income

Income group

ioth decileist and 2nd quintiles

distribution

1780

51-53IO-II

derived from Table 7

1831

4518

(per cent).

1866

4916

Note: Category: Labour force.Sources: Tables 6 and 7.

sons in the lowest ranks of the income scale was the large influx of ruralworkers who were happy to accept urban employment for wages that,although low, allowed them to benefit from the enhanced quality of lifeavailable in urban areas.

The information contained in Table 7 cannot be transformed into acomplete income distribution, but with only minor adjustments it can beused to estimate the extreme classes of an income distribution. Theseare presented in Table 8 where the similar information for 1780 isrepeated.

The data demonstrate clearly the decrease in inequality that occurredbetween 1780 and 1831, and the subsequent reversal. The degree ofinequality that existed in the years prior to the Revolution was neverafterwards attained. Although the greater equality that prevailed in 1831subsequently eroded, the income distribution in 1866 exhibited greaterequality than had prevailed prior to the Revolution.

One factor that served as a brake on the rise of inequality after 1831was the relatively strong increase in agricultural productivity that occurredthroughout most of the nineteenth century. Between 1831 and 1866 totalemployment in France increased by 21 per cent, but the number of per-sons in the agricultural sector remained constant. During the same periodthe volume of agricultural production increased by nearly 50 per cent(Toutain 1987, p. 98-102). This substantial increase in agricultural outputwas accomplished with no increase in agricultural labour, thus resulting inthe sharp increase in agricultural productivity without which there wouldhave been an even greater increase in inequality between 1831 and 1866.

4.2. The income distribution in 1894

In the last decade of the nineteenth century there was a flurry of attemptsto produce income distributions. Much like the primitive attempts of theeighteenth century, the efforts of the late nineteenth century were mostlyundertaken to illustrate the effects of various proposals for an income taxthat was eventually implemented in 1917.

Colin Clark (1951, p. 535) mistakenly attributed to Pareto an estimateof the income distribution for France in the 'Middle Ages' that he said


Table 9. The income distribution: 1894 and 1899-1901 (percentage oftotal income).

Income group Coste's Doumer's Caillaux's Colson'sestimate for estimate estimate estimatefor 1894 for 1894 for 1894 1899-1901

10th decile9th decile4th quintile3rd quintile2nd quintile1st quintile

Total

48.015.517.0IO.O6.0

3-5100.0

41.010.018.014.0IO.O

7.0100.0

42.0IO.O

18.513.0IO.O6.5

100.0

46.0IO.O17.012.0

9.06.0

100.0

Note: Category: Population.Sources: Coste (1895), Colson (1903 and 1927), and Morrisson (1991).

had been based on ' income from rents in Paris ' . Pareto had published adistribution based on Paris rents but it was for 1889, and he neverintended it to be interpreted as an income distribution.8 The first attemptto estimate an income distribution for France in the closing decade of thenineteenth century was completed by Adolphe Coste, a senior employee ofthe Ministry of Finance, who was responsible for collecting and analysingstatistics for the contribution personnelle-mobiliere tax for 1894.9 The tax hadmore or less replaced the Old Regime's capitation tax although the Ministryof Finance's data were not based on actual collections but rather on whatthe tax inspectors presumed was the real rental value of residences andother buildings. Since the tax was imposed on every unit in a building,virtually all family units were included except those boarding with others.However, the data used by Coste did not include Paris, a deficiency thatwas eventually corrected by others. Using these data, Coste (1895) madean estimate of the income distribution by presuming that the tax wasequivalent to 0.6 per cent of a person's income. He presented his results tothe National Assembly as part of the debate about introducing an incometax.

Over the next three decades the 1894 tax data were used by several othersthat including two ministers of finance, Doumer and Caillaux, as a basis forestimating the revenue that might be collected from several proposed

8 This study was first published in the Bulletin de Statistique et Legislation Comparee(September 1890), and later included in his first publication (1896) about the 'Paretocoefficient'. Later he published (1897) a second distribution for France, but it waslimited to analysing how the shares of the Credit Foncier de France for the years 1888and 1895 were distributed among the owners.

9 Earlier Coste (1890) had presented an estimate of the socioeconomic structure of Franceto the Societe de Statistique de Paris on 18 June 1890.


i oth decile9th decile4th quintile3rd quintile2nd quintileist quintile

TotalGini coefficients:

51-5311-15

168

JIO100

0.60/0.62

Table 10. Income distributions for IJ8O3 1864 and 1899-1901(percentage of total income)

Income group 1780 1866 1899-1901

48-49 4510

1712

6.5100 100

0.48

Note: Category: Population.Sources: Tables 6, 8 and 9.

income taxes.10 In 1903 the engineer cum economist Colson published arevision based on the 1894 tax data, and then 24 years later (1927) producedwhat is the most reliable estimate of an income distribution for the turn ofthe century (1899-1901). Between Colson's two estimates, there had beenanother revision of the 1894 data by Levasseur (1907). Colson's last esti-mate is presented in Table 9 where we also present Coste's original estimatefor 1894, remembering that this estimate was made before data for Pariswere available.

Doumer stated that he had purposely underestimated the income ofthose in the highest bracket in order not to create too optimistic a view ofhow much an income tax might produce. Therefore, it is perhaps surprisingthat his estimate for the tenth decile is not lower than shown. Also, recog-nizing that Coste did not have access to the tax data for Paris when he madehis first estimate, it is somewhat puzzling that his estimate for the tenthdecile is the highest among the four late-nineteenth century estimates. Evenso, the differences between the four estimates are not large, and except forCoste, the estimates for the bottom two quintiles are quite similar. In orderto evaluate the evolution of the income distribution from the late eighteenthcentury until the beginning of the twentieth century, we present the threeestimates we have discussed above in Table 10.

5. The Kuznets hypothesis

After analysing the evolution of the income distributions for severalindustrialised countries, Kuznets (1955) discerned a general pattern that has

10 The next modification of the 1894 t a x data was made by a new Minister of Finance,Doumer (Colson 1903, p. 313). Another revision of the data was made in 1907 by thenext Minister of Finance, Caillaux, again in support of his arguments for adopting anincome tax (Colson 1927, p. 419).

The income inequality of France in historical perspective JJ

subsequently been referred to as Kuznets' 'inverted U' hypothesis. TheKuznets hypothesis is that as a country begins the process of industrialdevelopment, income inequality increases but eventually greater in-come equality occurs, producing an inverse 'U' with respect to incomeinequality.

The income data for France are not sufficiently precise nor continuous toprovide a strong test of the Kuznets hypothesis, but what emerges from theavailable data is that the hypothesis seems to be essentially supported by thecase of French industrialisation. In the years just prior to the Revolution,income inequality was probably as high as it has ever been anywhere inmodern times. While there are no quantitative data about how that distri-bution may have changed in the decades immediately following theRevolution, there is considerable reason to believe that a substantialdecrease in inequality occurred, ending perhaps about 1830. This period ofdecreasing inequality occurred prior to the commencement of industrialisa-tion in France. If it is accepted that the industrial revolution in Francebegan sometime after 1830, then this was exactly the period during whichincome inequality must have begun to increase, because by 1866 we findincome inequality almost as high as it was prior to the Revolution. Thirtyyears later, and based on data that are more reliable than the data availableearlier, we find that income inequality had decreased significantly and at atime that is more than a half century after the beginning of the industrialrevolution in France. Thus, Kuznets' 'inverted U' hypothesis is further sup-ported by the experience of France.

However, two aspects of the evolution of the income distribution arenotable. First, the degree of inequality before the Revolution was alreadyhigh, even higher than in 1866, nearly a century later. Kuznets concludedfrom his research that countries generally had a lower degree ofinequality before beginning their phase of industrial revolution. Second, inFrance there was an important decline in inequality during theRevolutionary years 1790-1815. Again this is different from anythingKuznets had observed elsewhere; the decline resulted from France's uniqueRevolutionary experience.

6. The Theil coefficients

Although the income data have margins of error and we have no clearindications about how large the errors may be at any particular point oftime, it is useful to examine what they portray when their information isused to calculate Theil coefficients. The 'total' Theil coefficient can bedecomposed into two parts: the intra-group dispersion (variation amongmembers of the same group) and the inter-group dispersion (variationamong the groups). Except for the data extracted from the capitationrolls, we have no direct information about the magnitude of the intra-group


Year

178818661894

Table 11. Inter-group,

Inter-groupTheil

0.590.570.36

intra-groupj and total

Intra-groupTheil

0.360.360-33

Theil coefficients.a

TotalTheil

0.950.930.69

Notes: Category: Population. aThe formulas for the Theil coefficients are given in anAppendix.Sources: For the intergroup Theils: Tables 3, 7 and Morrison (1991). For the intergroupTheils: see text.

variations. However, knowledge about the members of each groupmakes it possible to advance some reasonable assumptions about thevariations:

(1) For the top decile the intra-group variation was undoubtedlylarge because the fortunes of the wealthiest members were manytimes those of the less wealthy members. For this group weassume an intra-group Theil of 0.5.

(2) Within each of the sub-groups of civil servants, small farmers,agricultural workers, day labourers, and servants there was littledispersion in incomes as the members of each group were quitehomogeneous. For these sub-groups we assume an intra-groupTheil of 0.1.

(3) The remaining groups that included the shopkeepers, artisansand large scale farmers exhibited differences in income thatwere probably about midway between the above two categories.For these we assume an intra-group Theil of 0.25.

Again, the validity of our assumptions is not known precisely enough toform strong conclusions based on the resulting Theil coefficients, but theyprovide a way of indicating a general order of magnitude, keeping in mindthe approximate nature of the underlying assumptions. The assumptionsmake it possible to combine the inter-group Theil estimates with theintra-group components in order to arrive at the total Theil coefficientsfor the socio-professional distribution of income that are presented inTable 11.

The composition of the Theil coefficients suggests that there was only amodest decrease in inequality within groups over the nearly hundred yearsbetween 1788 and 1894, but that the overall decrease between groups wassubstantial. The data also suggest that the degree of inequality in 1788 and1866 was nearly similar. And since we have reason to believe that inequalitydecreased after the Revolution, then it follows that inequality must haveincreased sometime afterwards. It is only by 1894 that we have reliable data


that inequality had decreased from the higher levels it exhibited in the eigh-teenth century and earlier in the nineteenth century as well.

Another analysis is possible with the Theil coefficients. We can calculatethe simple inter-sectorial Theil coefficients for the degree of inequalitybetween the two principal sectors of the economy: agriculture and non-agri-culture. These are as follows:

Table 12. Inter-sectorial Theil coefficients.

Year Sector N Y; Inter-sectorial(percentages) Theil (T )

1788

1864

1894

AgricultureNon-agricultureAgricultureNon-agricultureAgricultureNon-agriculture

683248524852

45554456406 0

O.II

O.OI

0.02

Note: Category: Population.Sources: Tables 3, 7 and Morrisson (1991).

This analysis highlights the originality of the French experience. Whereasthere was a certain degree of income inequality between the agriculturalsector and the rest of the economy during the eighteenth century, apparentlythe inequality was much diminished by the middle of the nineteenth centuryand remained so by the closing decades of that century. A consequence ofthis is that the reduction in inequality that occurred in France between 1866and 1894 cannot be attributed to a reduction in income inequality betweenthe agricultural and non-agricultural sectors because the degree of inequalitywas already small by 1866. The small difference in income inequalitybetween the agricultural and non-agricultural sectors in France during thenineteenth century is unusual compared to the experiences of otherEuropean countries during similar phases of their economic development.

7. Summary and conclusions

This analysis of the evolution of the income distribution for France islimited due to the lack of reliable statistics concerning the economy duringmuch of the period during the eighteenth and nineteenth centuries.Nevertheless, the results suggest that the degree of inequality that existed inthe later part of the eighteenth century, and undoubtedly earlier in the cen-tury also, was high and perhaps even higher than existed in England at acomparable period in its economic development.11

11 For analyses of the historical evolution of income inequality in England, see: Brown(1988)5 Lindert (1994^ 1999)5 and Williamson (1979, 1980, 1985).

8o European Review of Economic History

In the absence of firm statistical data concerning the years following theFrench Revolution, it is impossible to quantify precisely the change ininequality. However, several changes in economic conditions corroboratethe view that inequality decreased. How large was the decrease, how longdid it decline, and when did it begin to increase . . . these are all import-ant historical questions but ones for which our analysis does not provideclear answers. It does seem that the turnaround may have occurred some-time near 1830 when France's industrial revolution got underway. Whatseems more certain is that by 1866 the degree of inequality had increasedand was at a level nearly comparable to what had existed in the years priorto the Revolution.

Kuznets observed that the initial phase of economic development usuallycoincided with an increase in inequality. He attributed this to the combi-nation of a constant and significant income disparity between the agricul-tural and the non-agricultural sectors, and decreasing employment in theagricultural sector during the process of economic growth. In Francebetween 1830 and 1894, there is evidence that income inequality firstincreased and then began decreasing, confirming Kuznets' 'inverted U'.However, intersectoral inequality between agriculture and the other sectorscontributed virtually nothing to explaining the change in income inequalitybetween 1864 and 1894 (see the Theil coefficients, Table 12). France wasunusual because agricultural income increased rapidly between 1830 and1894 with the result that the income disparity between the agricultural andthe non-agricultural sectors decreased.

While not as comprehensive as data from the national accounting systemsthat countries began using in the second half of the twentieth century, thedata for the second half of the nineteenth century strongly suggest that asignificant decrease in inequality occurred between 1866 and 1894. Theabsolute degree of inequality was still high in 1864, and the decrease thatoccurred between 1866 and 1894 was significant. In the twentieth century,the degree of inequality has generally continued to decrease, although thedecline has not been without interruptions. Sauvy (1967) estimated that by1929 the bottom 40 per cent received 17 per cent as compared with 10 percent in the late eighteenth century. By 1985, after allowing for governmenttaxes and transfers, the share going to the bottom 40 per cent had risen to23 per cent. In the upper ranges of the income distribution, the decrease ininequality was equally notable. In 1890 the fifth quintile received 55 per centof income, and by 1985 it had decreased to 47 per cent (or 40 per cent aftertaxes and transfers); the tenth decile's income had decreased during thesame period from 45 per cent to 31 per cent (or 26 per cent after taxes andtransfers).12

It is evident that our estimates represent only a first step toward

12 See Morrisson (1991).


determining how income inequality evolved in France between the latterpart of the eighteenth century and the beginning of the twentieth century.The next steps to obtain greater precision would be to explore morethoroughly the capitation rolls and other taxes that were imposed during theeighteenth century, and especially investigating whether suitable proxies forincome can be found for the first half of the nineteenth century.Nevertheless, we believe our estimates provide a solid beginning forunderstanding the main changes that occurred during the two centuriescovered.by this study.

Acknowledgement

The authors benefited from the incisive comments of two anonymous referees andthe helpful suggestions of the editors of this review.

References

BLUCHE, F. and SOLNON, J-F. (1983). La Veritable Hierarchie Sociale de VAncienneFrance: Le Tarifde la Premiere Capitation. Geneva: Droz.

BoiSLJSLE, A. (ed.) (1874). Correspondance des Controleurs Generaux des Finances,vol. 1:1683-1689. Paris: Imprimerie Nationale.

BONNEYJ R. (ed.) (1995). Economic Systems and State Finance. Oxford: ClarendonPress.

BONNEY, R. (ed.) (1999). The Rise of the Federal State in Europe. Oxford: OxfordUniversity Press.

BRAUDEL, F. and LABROUSSE, E. (eds) (1970). Histoire Economique et Sociale de laFrance, part 2, vol 2: 1660—1789. Paris: Presses Universitaires de France.

BRAUDEL, F. and LABROUSSE, E. (eds) (1976). Histoire Economiques et Sociale de laFrance, part 3, vol. 1:1789-1880. Paris: Presses Universitaires de France.

BROWN, H. (1988). Egalitarianism and the Generation of Inequality. Oxford:Clarendon.

CHANUT, J - C , HEFFER, J., MAIRESSE, J. and POSTEL-VINAY, G. (1995). Lesdisparities de salaires en France au XIXe siecle. Histoire etMesure 10, pp. 381-409.

CLARK, C. (1951). The Conditions of Economic Progress. Paris.COLSON, C. (1903). Cours d'Economie Politique, vol. 2. Paris: Guillaumin.COLSON, C. (1927). Cours d'Economie Politique, vol. 3, 2nd edn. Paris: Guillaumin.COSTE, A. (1890). Etude Statistique sur les Salaires des Travailleurs et le Revenu de la

France. Paris: Guillaumin.COSTE, A. (1895). Rapport General presente au Norn de la Commission Extra

Parlementaire: Proces-Verbaux, vol. 2. Paris: Imprimerie Nationale.DES CILLEULS, A. (1885). La Population Frangaise avant 1789. Paris.DUMKE, R. (1991). Income inequality and industrialisation in Germany,

1850-1913; the Kuznets hypothesis re-examined. In Y. S. Brenner, H. Kaeble,and M. Thomas (eds.), Income Distribution in Historical Perspective. Cambridge:Cambridge University Press.

DUPAQUIER, J. (1991). Histoire de la Population Frangaise, vol. 2: De la Renaissancea 1789. Paris: Presses Universitaires de France.


EXPILLY, ABBE DE (1780). Tableau de la Population de la France. Amsterdam.FEINSTEIN, C. (1988). The rise and fall of the Williamson Curve. Journal of

Economic History 48, pp. 699-729.GUERYJ A. (1986). Etat, classification sociale et compromis sous Louis XIV: la

capitation de 1695. Annales ESC 41, pp. 1041-60.HuFTONj O. (1974). The Poor of 18th Century France. Oxford: Oxford University

Press.INSTITUTE NATIONAL DE LA STATISTIQUE ET DES ETUDES ECONOMIQUES, Serie

menages. Les revenus fiscaux des menages.ISNARDj A. (1781). Traite des richesses. Londres et Lausanne.KUZNETS, S. (1955). Economic Growth and Income Inequality. American

Economic Review 45, pp. 1-28.KWASS, M. (1994). Liberty, Egality, Fiscality: Taxation, Privilege and Political

Culture in Eighteenth-Century France. Ph.D. thesis, Ann Arbor, MI.LEVASSEUR. E. (1907). Questions Ouvrieres et Industrielles en France. Paris: Arthur

Rousseau.LlNDERT, P. H. (1980). English occupations, 1670-1811. Journal of Economic

History 40, pp. 685-712.LlNDERTj P. (1986). Unequal English wealth since 1670. Journal of Political

Economy 94, pp. 1127—62.LINDERT, P. (1994). Unequal living standards. In R. Floud and D. McCloskey,

(eds) The Economic History of Britain since 1700, vol 1:1700-1860. Cambridge:Cambridge University Press.

LINDERT, P. (1999). Three centuries of inequality in Britain and America. In T.Atkinson and F. Bourguignon, Handbook of Income Distribution. Amsterdam:North Holland.

LINDERT, P. H. and WILLIAMSON, J. G. (1983). Reinterpreting Britain's SocialTables, 1688—1913. Explorations in Economic History 20, pp. 94-109.

MARCHAND, O. and THELOT, C. (1997). Le Travail en France. Paris: Nathan.MARION, M. (1910). Les Impots Directs sous I'Ancien Regime. Paris: Corneli.MORRISSON, C. (1991). L'inegalite des revenus, in J.-Cl. Casanova and M. Levy-

Leboyer (eds), Entre I'Etat et le Marche. L'Economie Francaise des Annees 1880 anos Jours. Paris: Gallimard.

PARETO, A. (1890). Bulletin de Statistique et Legislation Comparee, September.PARETO, A. (1896). La Courbe de la Repartition de la Richesse. Geneve: Universite

de Lausanne Recueil, Faculte de Droit.PARETO, A. (1897). Aggiunta allo studio sulla curva della entrate. Giornale degli

Economist^ 2nd series, 14, pp. 15-26.PEUCHET, J. (1805). Statistique Elementaire de la France. Paris.POSTEL-VINAY, G. (1989). A la recherche de la revolution economique dans les

campagnes (1789-1815). Revue Economique 40, pp. 1015-45.RENARD, LE SIEUR. (1790). Decouverte de la Valeur Certaine du Produit Reel et

Absolu du Sol de la France. Paris.SAUVY, A. (1967). Histoire Economique de la France entre les Deux Guerres

(1931-1939)3 vol 2. Paris: Fayard.SNYDER, W. and OSTROOT, N. (forthcoming). Evolution de la structure socio-

professionnelle des villages du Sud-Est de la France: 1695-1866. ProvenceHistorique.


SOLTOW, L. (1987). The distribution of income in the United States in 1798:estimates based on the Federal Housing Inventory. Review of Economics andStatistics 69, pp. 181-5.

SOLTOW, L. (1989). Income and Wealth Inequality in Amsterdam, 1585-1805.Economisch-en-Sociaal-HistorischJarrboek, pp. 72-95.

THORILLON, A. (T. MINAU DE LA MISTRINGUE) (1787). Idees sur les Impots Publics.Paris.

TOUTAIN, J-C. (1987). Le Produit Interieur Brut de la France de 1789 a 1982. Paris:Cahiers de TI.S.M.E.A.

TOUTAIN, J-C. (1992). La Production Agricole de la France de 1810 a 1990:Departements et Regions, t.l. Paris: Cahiers de 1'ISEA, serie AF.

TOUZERY, M. (1994). L'Invention de VImpdt sur le Revenu: La Taille Tarifee1715-1789. Paris: Ministere de l'Economie: Comite pour l'Histoire Economiqueet Financiere de la France.

VERLEYj P. (1993). Domestic Demand and Industrial Growth in France (1830-1913).Cambridge, MA: National Bureau of Economic Research.

WEIR, D. (1991). Les crises economiques et les origines de la Revolutionfrancaise. Annales ESC 46, pp. 917—47.

WILLIAMSON, J. (1979). Two centuries of British income inequality:reconstructing the past from tax assessment data, 1708-1919. Economic HistoryDiscussion Paper Series. Maddison, WI: University of Wisconsin.

WILLIAMSON, J. (1980). Earnings Inequality in Nineteenth-Century Britain.Journal of Economic History 60, pp. 457-75.

WILLIAMSON, J. (1985). Did British Capitalism Breed Inequality? Boston: Allen andUnwin.

WORLD BANK (1998). World Development Report for 1998. New York: OxfordUniversity Press.

ZYLBERMAN, E. (1969). La Croissance et les Comptes Economiques de la France auSecond Empire. Paris: Faculte de Droit et des Sciences Economiques.

Appendix: Formulas for Theil coefficients

Formulas for Theil coefficients where Y; = percentage income of the rth group, N= percentage of population in the *th group, and T = the assumed Theil for thedispersion of income within the zth group.

Inter-group Theil: Ta = 2[(In J ^

Intra-group Theil: Tb = XYTj

Total Theil = Inter-group Theil (Ta) + Intra-group Theil (Tb).

the income inequality of france in historical...

Documents